Prospective Hyspiri Global Observations of Tidal Wetlands

Remote Sensing of Environment 167 (2015) 206–217

Contents lists available at ScienceDirect

Remote Sensing of Environment

journal homepage: www.elsevier.com/locate/rse

Prospective HyspIRI global observations of tidal wetlands

Kevin R. Turpie a,⁎, Victor V. Klemas b, Kristin Byrd c,MaggiKellyd,Young-HeonJoe a Joint Center for Earth Systems Technology, University of Maryland, Baltimore County, 5523 Research Park DR | #320, 21228 Baltimore, MD, USA b School of Marine Science and Policy, University of Delaware, Newark, DE 19716, USA c Western Geographic Science Center, U.S. Geological Survey, 345 Middleﬁeld Road, Menlo Park, CA 94025, USA d Department of Environmental Sciences, Policy and Management, University of California, Berkeley, 137 Mulford Hall, #3114, Berkeley, CA 94720, USA e Department of Oceanography, Pusan National University, Busan 609-735, South Korea article info abstract

Article history: Tidal wetlands are highly productive and act as critical habitat for a wide variety of plants, fish, shellfish, and Received 17 June 2014 other wildlife. These ecotones between aquatic and terrestrial environments also provide protection from Received in revised form 11 May 2015 storm damage, run-off filtering, and recharge of aquifers. Many wetlands along coasts have been exposed to Accepted 15 May 2015 stress-inducing alterations globally, including dredge and fill operations, hydrologic modifications, pollutants, Available online 16 June 2015 impoundments, fragmentation by roads/ditches, and sea level rise. For wetland protection and sensible coastal development, there is a need to monitor these ecosystems at global and regional scales. Recent advances in sat- Keywords: Wetlands ellite sensor design and data analysis are providing practical methods for monitoring natural and man-made HyspIRI changes in wetlands. However, available satellite remote sensors have been limited to mapping primarily Satellite remote sensing wetland location and extent. This paper describes how the HyspIRI hyperspectral and thermal infrared sensors Hyperspectral can be used to study and map key ecological properties, such as species composition, biomass, hydrology, and Imaging spectroscopy evapotranspiration of tidal salt and brackish marshes and mangroves, and perhaps other major wetland types, Thermal imaging including freshwater marshes and wooded/shrub wetlands. Remote sensing © 2015 Published by Elsevier Inc. Biogeography Coastal science

1. Introduction improvement through filtering of agricultural and industrial waste, and recharge of aquifers (Barbier et al., 2011, 2008; Odum, 1993). The Hyperspectral Infrared Imager (HyspIRI) will likely offer unique These ecosystems have been impacted by a wide range of stress- opportunities to globally study ecosystems where land meets sea. These inducing alterations, including dredge and fill operations, hydrologic environments play a critical role in global biogeochemical cycles and in modifications, pollutant run-off, eutrophication, impoundments, shaping and sustaining marine and terrestrial ecosystems. They are invasion by non-native plant species, and fragmentation by roads and highly complex, interconnected, and are important to human habitats ditches (Zedler & Kercher, 2005). There is also considerable concern re- and economy. They are changing with climate and increasing human garding the impact of climate change on coastal wetlands, especially activities. Remote sensing is a critical tool for the study of inland and due to relative sea level rise, increasing temperatures and changes in coastal waters and their watersheds and the advent of hyperspectral precipitation (Church & White, 2006; McInnes, Walsh, Hubbert, & imaging is expected to be transformational to that capability. However, Beer, 2003; Watson et al., 2014). Increased CO2 levels and temperature HyspIRI will not only provide global 30-meter resolution hyperspectral changes may change the current distribution of C3 and C4 plants imagery, it will also provide contemporaneous thermal data at (Bromberg-Gedan, Silliman, & Bertness, 2009). Most marshland grass 60-meter resolution. This is an unprecedented opportunity and offered species are C4 plants and are not expected to increase productivity by no other planned space mission, domestic or international. In this under higher CO2 levels, but that is probably not the case for C3 plants paper, we discuss the benefits and challenges of HyspIRI observations (Erickson, Megonigal, Peresta, & Drake, 2007), which include many of to the global study of coastal tidal wetlands. the marsh sedges and rushes. Rising temperatures from global warming Tidal wetlands and estuaries are highly productive and provide may bring invasion of warmer water species into colder zones (Zomer, important habitat for a wide variety of plants, fish, shellfish, and other Trabucco, & Ustin, 2009). At the same time, tidal ecosystems (including wildlife (Batzer & Baldwin, 2012). Tidal wetlands also provide flood marshes, mangroves, and seagrasses) represent a significant standing protection, a buffer from storm and wave damage, water quality carbon pool in soils and vegetation. The carbon stocks and future cumu- lative carbon storage in these systems are referred to as “Blue Carbon,” ⁎ Corresponding author. and play an important role in managing atmospheric carbon E-mail address: [email protected] (K.R. Turpie). (Pendleton et al., 2012). In an extensive study by Cebrian (2002), tidal

http://dx.doi.org/10.1016/j.rse.2015.05.008 0034-4257/© 2015 Published by Elsevier Inc. K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217 207 marshes and mangroves were also shown to have the greatest carbon 2010; Pengra, Johnston, & Loveland, 2007; Schmidt et al., 2004; Ustin, production and flux per area compared to other types of terrestrial Roberts, Gamon, Asner, & Green, 2004; Yang, Everitt, Fletcher, Jensen, and aquatic coastal ecosystems. Thus changes in these systems could & Mausel, 2009; Zomer et al., 2009). In addition with hyperspectral, also influence the role of coasts in biogeochemical cycles or affect the narrow-band vegetation indices, researchers have made progress on status of coastal carbon reservoirs. estimating biochemical and biophysical parameters of wetland vegeta- To plan for protection and responsible development of these tion, such as water content, biomass and leaf area index, and leaf wetlands, there is a need to map and monitor changes in tidal wetlands constituents such as nitrogen, as demonstrated in the literature (see at local, regional and global scales (Ozesmi & Bauer, 2002). Much has Adam et al., 2010; Artigas & Yang, 2006; Filippi & Jensen, 2006; been done since the early 1980s to map wetlands nationally with re- Gilmore et al., 2008; O'Connell, Byrd, & Kelly, 2014; Ozesmi & Bauer, mote sensing particularly through the U.S. Fish and Wildlife Service 2002; Pengra et al., 2007; Tian et al., 2011; Wang, 2010). (USFWS) National Wetlands Inventory (NWI) (Dahl & Stedman, 2013) Hyperspectral classification is an important analysis tool for study- and the NOAA Coastal Change Analysis Program (NOAA, 2015)and ing salt and brackish marsh systems. Tidal salt and brackish marshes mapping wetlands continues to be a national priority in the USA, and tend to form monospecific canopies, with certain species favoring global efforts to monitor wetlands are increasing (USFWS, 2015). Early different salinity and drainage conditions that change geographically. work with multispectral imagery has been based predominantly on As a result, the landscape becomes partitioned into zones of plant Landsat Thematic Mapper data (Klemas, 2011, 2013a,b). This legacy species that can deal with different levels of salinity, a process called continues with the recently launched Landsat 8 Operational Land zonation (Adam, 1990; Day et al., 2012). Observing changes in zonation Imager (OLI), which was used to derive the color-infrared (CIR) image can provide information about how a marsh system is responding to in Fig. 1, an example of a tidal wetland system on the eastern shore of changes in hydrologic processes or sea level. Change in distribution Chesapeake Bay. can also be caused by invasive species. Hyperspectral classification tech- More recently, wetland mapping has been augmented with the niques have been applied to map the distribution of marsh species use of hyperspectral imagery (Adam, Mutanga, & Rugege, 2010). (Adam & Mutanga, 2009; Adam et al., 2010; Andrew & Ustin, 2008; Hyperspectral remote sensing data have been shown to improve efforts Artigas & Yang, 2005; Barducci, Guzzi, Marcoionni, & Pippi, 2009; to map wetland location and extent, classify wetlands according to Judd, Steinberg, Shaughnessy, & Crawford, 2007; Zomer et al., 2009). vegetation functional types or species, and derive critical functional Unlike multi-band sensors, hyperspectral imagers can detect fine differ- properties of wetlands across types and scales (Brando & Dekker, ences in spectral features, enhancing wetland species discrimination 2003; Christian & Krishnayya, 2009; Hirano, Madden, & Welch, 2003; (Brando & Dekker, 2003; Christian & Krishnayya, 2009; Papes et al., Jensen et al., 2007; Papes, Tupayachi, Martínez, Peterson, & Powell, 2010; Pengra et al., 2007).

Fig. 1. Satellite imagery of coastal wetlands. Shown is part of a Landsat 8 Operational Land Imager (OLI) Color-Near IR (CIR) scene over the western coast of the Delmarva peninsula, south of Cambridge, Maryland USA on 2 February 2014. The region corresponds to the map depicted in Fig. 2. Melting remnants of recent snowfall can be seen as small light blue or white spots, which suggests increased flow rates and sediment fluxes. In this region, residential and agricultural land use at higher elevations of the watershed feed nutrients and soil into the wetlands below. What the wetlands do not filter into the adjacent estuaries. In the east is the Chesapeake Bay and to the right are Fishing Bay, Nanticoke River, and Money Bay. The confluence of the last three connects with the Chesapeake via Hooper's Strait between Bishops Head and Bloodworth Island. Fresher, more soil and CDOM rich waters mix with the “greener” Chesapeake waters just east of the strait. The waters in the northern end of Fishing Bay, and in Blackwater Lake further northwest, are dark with heavy concentrations of chromophoric dissolved organic matter (CDOM), a product of decaying vegetation. The red–green–blue (RGB) imagery inset also shows the appearance in these estuaries of heavy concentrations of suspended sediment, which only observable by including the red channel. High-resolution spectroscopy and thermal measurements from HyspIRI provide further information regarding the biological and environmental responses to these processes. Credits: USGS. 208 K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217

In the following sections, we focus our discussion on coastal tidal Table 1 marshes and mangrove vegetation and processes involving hydrology Key characteristics of the Hyperspectral Infrared Imager (HyspIRI) mission. and hydrometeorology. The paper highlights areas in which the VSWIR TIR hyperspectral capability, spatial resolution, swath-width and signal- Spectral range 380 to 2500 nm 3.98, 7.35, 8.28, 9.07,10.53, to-noise (SNR) ratio of HyspIRI meet most of the requirements for 11.33, and 12.05 μm mapping wetland types and processes at regional to global scales. This Spectral bandwidth 10 nm, uniform over range 0.084, 0.32, 0.34, 0.35,0.36, summary is then followed with a review of specific HyspIRI data appli- 0.54, 0.54, and 0.52 μm cations for these various wetland types and processes. Finally the paper Radiometric resolution 14 bit 14 bit Angular field of view 12° 51° describes expected challenges associated with coastal wetland remote Altitude 700 km 700 km sensing. Swath width 185 km 600 km Cross track samples N2400 10,000 2. Hyperspectral sensors Spatial resolution 30 m (depth b 50 m) 60 m (depth b 50 m) 1 km (depth N 50 m) 1 km (depth N 50 m) Orbit Polar ascending Polar ascending Hyperspectral sensors have been primarily available on aircraft Equatorial crossing 11:00 a.m. 11:00 a.m. (Fearns, Klonowski, Babcock, England, & Phillips, 2011; Lesser & Equatorial revisit 16 days 5 days Mobley, 2007; Li, Ustin, & Lay, 2005; Ozesmi & Bauer, 2002; Rosso, Rapid response 3 days 3 days Ustin, & Hastings, 2005). However, to map and monitor wetlands Tilt 4° West 4° West globally, spaceborne platforms are required, and it is ideal if the data are available to public at no cost. To date, only three hyperspectral imagers have been placed in orbit. All three were intended to be low cost, prototype sensors that were developed as a proof-of-concept HyspIRI instrument to other existing and future sensors can be found with on-demand image acquisition of selected sites on the globe. in Table 2. None were designed to provide global wetland coverage on a regular Both PRISMA and EnMap have capabilities that would be useful to basis. study specific marsh systems, but neither are intended to produce glob- In 2000, the National Aeronautics and Space Agency (NASA) of the al or synoptic maps and neither have thermal bands. Thus, unlike USA launched the first spaceborne hyperspectral sensor, Hyperion HyspIRI, these missions cannot be expected to support global observa- aboard the Earth Observatory 1 (EO-1), with 220 contiguous spectral tions of tidal wetlands (e.g., to understand the effects of global climate bands. Images over selected sites covered a narrow strip (7.5 km by change), nor can they provide as extensive suite of contemporaneous 200 km) with 34 m spatial resolution. At the time of writing this paper, data products that HyspIRI offers, such as soil moisture or evapotranspi- NASA extended the mission one more year. However, it is highly unlikely ration, as discussed later in this paper. Also, the greater temporal resolu- that Hyperion will be in orbit during any date chosen for a HyspIRI tion of PACE and GeoCAPE could certainly complement HyspIRI launch. In 2001, the European Space Agency (ESA) launched the Com- observations. However, the respective spatial resolutions for PACE and pact High Resolution Imaging Spectrometer (CHRIS), aboard the Project GeoCAPE of 1000 and 250 m are far less capable of observing wetland for On-Board Autonomy 1 (Proba-1). CHRIS/Proba is a programmable structure. Also, only HyspIRI offers thermal bands, which can be useful sensor that can acquire 13 km × 13 km images of specified locations at in coastal wetlands studies. five viewing angles. The sensor has built-in modes for acquiring these To illustrate the importance of spatial resolution in observing tidal images with 17 m to 34 m spatial resolution, for 18 to 62 bands over a wetlands, a pixelization simulation was applied to a wetland mask to spectral range from 400 nm to 1050 nm (Barnsley, Settle, Cutter, Lobb, derive the percentage of pixel mixing as a function of resolution for a & Teston, 2004). The USA Naval Research Laboratory later built the large marsh system. Fig. 2 shows a map of wetland cover types from HyperspectralImagerfortheCoastalOcean(HICO),whichwassituated the United States Fish and Wildlife Service (USFWS) National Wetlands on the International Space Station (ISS) in 2009. HICO image dimensions Inventory (NWI). The area shown includes a large marsh system along were 42 km × 192 km, at about 90 m spatial resolution, for 128 bands the eastern shore of the Chesapeake Bay (USFWS, 2014). Polygons from 352 nm to 1080 nm. Because HICO was designed to observe coastal from a NWI shape file were used to create a layer in GIS software called waters, its radiometric performance was superior to Hyperion and QGIS and rasterized to approximate one-meter resolution. This cap- CHRIS/Proba in sensitivity and SNR. Unlike Hyperion and CHRIS/Proba, tured much of the small-scale features seen in the NWI polygon data, in- HICOcouldnotacquireimageryabove45°latitudebecauseoftheISSor- cluding variation along the marsh boundaries and within its interior, bital characteristics. Unfortunately, HICO operations ended due to an X- including small ponds and channels. The resulting image, with about class solar storm in September 2014 (see http://hico.coas.oregonstate. 104 pixels on a side, was loaded into IDL, which is software designed edu/). for image analysis. An array of the same dimensions as the NWI image In the next few years, Italy plans to launch the PRecursore was created. Elements of this array were set to unity where image pixels IperSpettrale della Missione Applicativa (PRISMA), which is a push were designated as being the Estuarine and Marine Wetland cover type broom imager with 30 m spatial resolution and a 30 km swath width. (see Fig. 2) and the rest of the array was left as zero. Then the array was PRISMA has over two hundred bands spanning 400 nm to 2500 nm. subsequently aggregated to increasingly coarser pixels sizes. The Germany plans to launch its Environmental Mapping and Analysis resulting average of ones and zeros in each aggregated pixel thus indi- Program (EnMap), which is designed to acquire a 30 km × 30 km cated the fraction of the selected wetland type that it contained. Fig. 3 image at 30 m spatial resolution, with 245 bands going from 420 nm uses a gray scale to illustrate the spatial distribution of this fraction to 2450 nm. within a magnified rectangular inset (identified by the small red box After Hyperion and HICO, the only hyperspectral missions planned in Fig. 2), which includes an area of marsh between the Blackwater by the USA include the Hyperspectral Infrared Imager (HyspIRI), the Lake in its northwestern corner and Fishing Bay in its southeastern cor- Pre-Aerosol, Clouds and Ecosystems (PACE) mission, and the Geosta- ner. The meandering river running between these two bodies is the tionary Coastal and Air Pollution Events (GeoCAPE) mission. HyspIRI Blackwater River, which becomes noticeably gray at as fine as 60 m res- will include an imaging spectrometer measuring from the visible to olution. The top row of simulated images represents the effects of shortwave infrared wavelengths in 10 nm contiguous bands, as well moderate-resolution sensors (30–90 m) and the bottom low- as a multispectral imager measuring from 3 μmto12μm in the mid- resolution sensors (e.g., PACE and GeoCAPE). Fig. 4 shows the fraction wave and thermal infrared wavelengths. Details regarding the mission of the wetland type that remains unmixed as the pixel size increases. and sensor characteristics are given in Table 1 and a comparison of the Generally, 30 m (the fourth circle) is where significant degradation K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217 209

Table 2 Feature comparison with existing and upcoming remote sensing resources.

Sensor Status Resolution Swath dims (km) Global? Source

Spatial (m) No. of bands Rep cycle (days)

Moderate spatial resolution HyspIRI — VSWIR Under development 30 212 16 185 Y USA HyspIRI — TIR Under development 60 8 5 600 Y USA Hyperion Termination planned 34 220 16 7.5 N USA Landsat OLI — VSWIR Operational 30 8 16 185 Y USA Landsat OLI — TIR Operational 100 2 16 185 Y USA EnMap Launch 2017 30 245 27 30 N Germany PRISMA Launch 2017 30 237 29 30 N Italy Sentinel 2 Launch 2015 (1st Sat) 10–60 13 5 290 Y ESA CHRIS/Probab Operational 17 18 7 13.5 N ESA ASTER — VSWIR Operational 15 9 16 60 N USA/Japan ASTER — Thermal Operational 90 5 16 60 N USA/Japan HISUI Launch 2018 30 185 60 30 Y Japan HICOa Terminated 90 87 3 42 N USA Low spatial resolution PACE — VisNIR Under development 500–1000 110 2 2300 Y USA PACE — SWIR Under development 500–1000 3 2 2300 Y USA GeoCAPE — VisNIR Under development 250–375 N1000 N0.04 300–500 N USA GeoCAPE — SWIR Under development 250–375 3 0.1 300–500 N USA GOCI Operational 500 8 0.04 2500 N Korea MODIS — VSWIR Operational 250–1000 21 2 2300 Y USA MODIS — Thermal Operational 1000 15 2 2300 Y USA VIIRS — VSWIR Operational 250–750 14 b2 3000 Y USA VIIRS — Thermal Operational 250–750 7 b2 3000 Y USA MISR Operational 275–1100 4 9 360 Y USA OLCI/Sentinel 3 Launch 2016 300 21 2 1270 Y ESA High spatial resolution SPOT-5 — VSWIR Operational 10–20 3 5 120 N Commercial SPOT-6 — VisNIR Operational 6 4 5 60 N Commercial SPOT-7 — VisNIR Operational 6 4 5 60 N Commercial QuickBird IIa Terminated 2.6 4 4 18 N Commercial IKONOS Operational 4 4 14 13–70 N Commercial Pleiades Operational 0.5 4 1 20 N Commercial GeoEYE 1 Operational 2 4 2 50 N Commercial RapidEye Operational 1.8 4 1 100 N Commercial Worldview 2 Operational 2 8 4 16 N Commercial Worldview 3 — VSWIR Operational 1.2 28 b1 13 N Commercial

VisNIR = 400–1000 nm; VSWIR = 400–2500 nm; TIR = 3–12 μm. a Terminated mission. b Set to mode (3b) for dark targets and highest spatial resolution. begins to occur and beyond 60 m the loss is substantial. Therefore, low- species and vegetation functional types and provide a better chance at resolution sensors are less appropriate for mapping, assessing, or observing the spectral signatures of plant stress and changes in produc- monitoring marshes, or similar wetland systems. tion. With its spectral range extending into the SWIR, the HyspIRI It should be noted that HyspIRI was originally planned to provide mission should provide broader use than HICO could in covering both imagery with global coverage at 60 m resolution, a swath width of aquatic and terrestrial features. 145 km, and equatorial revisit period of 19 days. However, in response HyspIRI's eight thermal bands will be more than adequate to observe to a congressional mandate in 2014 that the USA federal government subtle variations in water surface temperature, which would be ideal for should support Sustainable Land Imaging (SLI) for the next twenty- tracing fresh or melt water fluxes along water body margins with land five years, NASA instructed the HyspIRI mission management to and ice, respectively. Its VSWIR spectral capabilities could then observe conduct a study that would determine whether the HyspIRI mission aquatic biospheric response to contemporaneously observe coastal could supplement the existing Landsat imaging capability. A review of fluxes. The combined use of 30 m resolution satellite data from available technology led by the Jet Propulsion Laboratory (JPL) conclud- HyspIRI's optical and thermal infrared radiometers is promising for ed that replacing the originally proposed Offner spectrometer with a multiple applications, including the retrieval of latent and sensible Dyson spectrometer for HyspIRI would support Landsat spatial resolu- heat fluxes, and soil surface moisture variations (Petropoulos, Carlson, tion, while keeping the original spectral and radiometric characteristics. Wooster, & Islam, 2009; Sandholt, Rasmussen, & Andersen, 2002). This Therefore, as of this writing, HyspIRI mission will employ a VSWIR information will also be contemporaneous to the water and vegetation spectrometer that provides a 30 m spatial resolution equal to Landsat attributes derived from the hyperspectral measurements. legacy instruments. In addition, the new spectrometer design will have Landsat's 185 km swath width, which should also decrease the 3. HyspIRI and tidal wetland vegetation observations equatorial revisit period to 16 days (Green, 2015). With SNR comparable or better than the now defunct HICO instru- 3.1. Tidal marsh vegetation ment, HyspIRI should be more suited than Landsat to observe water optical characteristics (Devred et al., 2013), especially along shallow Coastal tidal marshes are the dominant coastal ecosystem in the coastal waters. This is useful because wetlands tend to be darker than United States and all temperate latitudes. They cover approximately typical terrestrial targets and it would also be useful to study the charac- 22,000 km2 of North America (Bridgham, Megonigal, Keller, Bliss, & teristics of adjacent water ecosystems and environments. This also Trettin, 2006). Tidal marshes are typically characterized by emergent could better support studies of interaction between coastal wetlands vegetation, and may have mineral or peat substrate. Tidal marshes are and estuarine and marine waters. HyspIRI's greater spectral resolution highly productive (Rocha & Goulden, 2009), and emergent plant growth and coverage should afford it superior capabilities in discriminating leads to peat formation in some of the most significant coastal areas of 210 K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217

Fig. 2. Major wetland types classified in the US Fish and Wildlife Service National (USFWS) Wetland Inventory (NWI) mapped over a portion of the eastern shore of the Chesapeake Bay in Maryland, USA. The area shown contains large a brackish marsh system that were used in a pixelization simulations. the United States, including South Florida, the Mississippi River Delta, et al., 2011). Studies of changes in ecological function and response and the Sacramento–San Joaquin River Delta (Miller & Fujii, 2010; are often limited to a small number of plots, and scientists must extrap- Nungesser, 2011; Tornqvist et al., 2008). olate findings to regional scales. Although monitoring widespread Salt marshes are widely distributed along many of the world's coasts changes to these landscapes could assist researchers and policymakers and are often dominated by specifically adapted species of grasses, in assessing and monitoring marsh deterioration or restoration, limited sedges and rushes that tend to form monospecific canopies across re- accessibility makes large-scale, in situ, evaluation challenging (Seher & gions with a high salinity gradient (Chapman, 1960). For instance, Tueller, 1973). Remote sensing techniques offer an efficient approach smooth cordgrass (Spartina alterniflora) tends to dominate the eastern to quantify changes in marsh vegetation (Klemas, 2013a,b). The utility North American seaboard and the Gulf of Mexico. The relative mono- of remote sensing techniques has been explored for measuring quanti- specificity and size of some salt marshes and aquatic nature of the sub- ties over large regions of wetlands, such as species and cover type strate make it possible to map them from satellites. Multispectral (Artigas & Yang, 2005; Jensen et al., 1986; Jollineau & Howarth, 2008; medium resolution satellite imagers, such as Landsat TM and Satellite Judd et al., 2007; Schmidt & Skidmore, 2003; Silvestri, Marani, & Pour l'Observation de la Terre (SPOT), and high resolution satellites, Marani, 2003; Underwood et al., 2006; Zomer et al., 2009), canopy such as IKONOS® and QuickBird, have been effective primarily for density or Leaf Area Index (LAI) (Sone, Saito, & Futakuchi, 2009; mapping wetland location and extent, at local or regional scales Wang, Huang, Tang, & Wang, 2007; Xavier & Vettorazzi, 2004; Xiao (Gilmore et al., 2010; Jensen et al., 1998; Klemas, 2011; Lunetta & et al., 2002), biomass (Byrd, O'Connell, Di Tommaso, & Kelly, 2014; Balogh, 1999; Lyon & McCarthy, 1995; Wang, 2010), as well as vegeta- Klemas, 2013a; Mishra et al., 2012; Mutanga, Adam, & Cho, 2012), or tion pattern and condition (Kelly, Tuxen, & Stralberg, 2011; Ramsey & quantities related to plant production and stress (Klemas, 2013a; Rangoonwala, 2005; Tuxen, Schile, Kelly, & Siegel, 2008; Tuxen et al., Mendelssohn, McKee, & Kong, 2001; O'Connell et al., 2014; Ramsey & 2011). Rangoonwala, 2006; Tilley, Ahmed, Son, & Badrinarayanan, 2003; Remote sensing efforts have been made to assess and monitor Vaesen, Gilliams, Nackaerts, & Coppin, 2001; Zhao et al., 2009). coastal marshes in order to improve our understanding of their essential Hyperspectral imagery, in its broad spectral range and radiometric services, as described above, and to aid in their management (Dahl & sensitivity could assist in improving applications in wetlands. Recent Stedman, 2013; Kelly & Tuxen, 2009; UNEP, 2006). Part of the process analysis of full spectrum field spectrometer data indicates that of managing degradation of coastal marsh services includes identifying hyperspectral first derivative reflectance spectra improve predictions changes in marsh systems that would affect these services (Barbier of wetland vegetation biomass over simulated broadband spectra K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217 211

Fig. 3. Effects of increasingly coarser resolution on spatial representation of an example wetland. Shown is an inset of a rasterized layer from polygon data mapped in Fig. 2 for the Estuarine and Marine Wetland type (see the double-line box in Fig. 2). under low inundations conditions (Byrd et al., 2014) or wetland canopy and also address atmospheric correction along coastal boundaries (Li structure (Fan, Xu, Liu, & Cui, 2010). Furthermore, biomass studies of et al., 2005; Ozesmi & Bauer, 2002; Rosso et al., 2005; Schmidt et al., grasslands demonstrate the greater capacity of hyperspectral imagery 2004). Detailed mapping of species composition, dynamics and plant to reduce the saturation problems in biomass estimation compared to vigor in complex salt marshes will require additional spectral data be- multispectral sensors (Mutanga & Skidmore, 2004), suggesting that yond what is found in multispectral imagery alone (Kelly & Tuxen, hyperspectral data will improve biophysical models in marshes as 2009). For example, AVIRIS (Airborne Visible and Infrared Imaging well, which are typically dominated by grass species like S. alterniﬂora. Spectrometer) has been successfully used to map estuarine wetland ex- Hyperspectral resolution over a broad spectral range is necessary to tent and type in the San Francisco Bay Area, California, USA (Li et al., discriminate and unmix cover types in these spectrally complex regions, 2005; Rosso et al., 2005), and Everglades National Park, Florida, USA (Hirano et al., 2003), as well as other study areas. These earlier proof- of-concept studies using airborne hyperspectral imagers demonstrate good potential utility for data generated by HyspIRI.

3.2. Mangroves

Mangroves help reduce the erosional impact of storms, serve as breeding and feeding grounds for juvenile fish and shellfish, trap silt that could smother offshore coral reefs and cleanse water by the uptake of nutrients and pollutants (Barbier, 2011; Ewel, Twilley, & Ong, 1998). Mangroves, which once occupied 75% of tropical and subtropical coast- lines, are now seriously threatened by coastal development and climate change, including sea level rise (Friess & Webb, 2014). In many coun- tries, mangrove swamps are being cut to provide firewood or building material and are being destroyed by development of shrimp ponds (Alongi, 2002; Heumann, 2011; Pinet, 2009; Tomlinson, 1995; Vaiphasa, Ongsomwang, Vaiphasa, & Skidmore, 2005; Wang & Sousa, 2009). Rapid losses of mangroves make it crucial to inventory and monitor the remaining mangroves to protect them from harmful development (Blasco, Aizpuru, & Gers, 2001; Gunawardena & Rowan, 2005; Terchunian et al., 1986). Mapping and quantifying the structure and biomass of mangrove ecosystems on a large scale is also important for Fig. 4. Effects of increasingly coarser resolution on pixel mixing over an example wetland. Shown faction of the Estuarine and Marine Wetland (simulated at one-meter resolution) studies of carbon storage, biodiversity, forest quality, and habitat suit- that remains unmixed as a function of pixel size. ability. Remote sensing has had a crucial role in monitoring mangroves, 212 K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217 but the majority of applications have been limited to mapping areal evapotranspired from land surfaces and 3% evaporated from open extent and patterns of change (Alongi, 2002; Blasco, Aizpuru, & Din water (Rivas & Caselles, 2004). In some zones of the world, about Ndongo, 2005; Heumann, 2011). 90% of the precipitation can be evapotranspired (Varni, Usunoff, Mangrove forests are patchy and can have gaps in the canopy that Weinzettel, & Rivas, 1999). Thus the greatest amount of water from expose moist soil or water. Therefore, high spatial resolution (b10 m) the hydrologic system is transpired and evaporated, showing the is required for mapping them in detail. Large-scale mangrove mapping importance evapotranspiration. has been performed in the past using medium-resolution satellites, Knowledge of latent and sensible heat fluxes, as well as soil water such as Landsat-TM and SPOT (Gao, 1998; Kovacs, Wang, & content, is important to many environmental applications, including Blanco-Correa, 2001; Saito, Bellan, Al-Habshi, Aizpuru, & Blasco, 2003). monitoring plant water requirements, plant growth, irrigation, land deg- SPOT multispectral data was shown to be suitable for mapping dense radation and desertification. Such data are also significant in the numer- mangroves. On the other hand, sparse mangroves were less accurately ical modeling and prediction of atmospheric and hydrologic cycles, and mapped, due to the spectral interference of their mudflat background for improving the accuracy of weather forecast models. The combined (Gao, 1998). use of satellite data from optical and thermal infrared radiometers has The advantages and problems associated with hyperspectral shown promise for the retrieval of latent and sensible heat fluxes and mapping have been clearly demonstrated by Hirano et al. (2003),who soil surface moisture variations within the top 5 cm of the soil depth used hyperspectral data to map vegetation for a portion of the Everglades (Ghilain et al., 2011; Marshall et al., 2013; Petropoulos et al., 2009). National Park in Florida. The hyperspectral imagers offered some opera- HyspIRI's thermal bands would provide additional information re- tional challenges, including large data volumes and more complex image garding wetland hydrology and hydro-meteorology. The evaporation processing procedures. However, the hyperspectral data proved effective flux from freshwater wetlands has been estimated using thermal in discriminating spectral differences among major Everglades species, infrared sensing data and a parameterization of the surface energy bal- such as red, black and white mangrove communities and enabled the de- ance (Jackson, 1985; Mohamed, Bastiaanssen, & Savenije, 2004). For in- tection of exotic invasive species. Furthermore, techniques for optimiz- stance, in the upper Nile Sudd wetlands of Sudan, the spatially averaged ing hyperspectral band selection for mangrove mapping have been evaporation over three years was found to vary between 1460 and developed (Koedsin & Vaiphasa, 2013). Laboratory experiments have 1935 mm/yr. This is substantially less than open water evaporation, also shown that discrimination between multiple species is possible and the wetland appears to be 70% larger than was previously assumed. (Heumann, 2011; Vaiphasa et al., 2005). Satellite-borne hyperspectral This new set of spatially distributed evaporation parameters from imagers, such as Hyperion, have been able to detect fine differences in thermal IR remote sensing forms an important dataset for calibrating a spectral reflectance. (Blasco et al., 2005; Heumann, 2011). With its regional climate model enclosing the Nile Basin (Chen, Kan, Tan, & high signal-to-noise ratio and good spatial resolution, the HyspIRI Shih, 2002; Mohamed et al., 2004). Having evaporation flux and spectral hyperspectral imager should be able to map mangrove extent and pro- types contemporaneous and collocated would better inform studies of vide major species discrimination on a global scale. ecosystem and canopy dynamics by enabling coordinated mapping of wetland extent and vegetation characteristics such as leaf area with hy- 3.3. Invasive species in wetlands drological processes. Spatial patterns of evapotranspiration in marshes can now be calcu- Wetlands can be invaded by plant species that displace native plants lated using satellite data with a minimum of ground meteorological and degrade habitat. Mapping tools are needed to document the loca- data (Meijerink, 2002). Rivas and Caselles (2004) have shown that the tion and extent of such invasive species. Pengra et al. (2007) were evapotranspiration of vegetation in a large river basin can be estimated able to map Phragmites australis, a tall grass that invades coastal by combining the surface temperature, as obtained from satellite im- marshes throughout North America, with Hyperion hyperspectral satel- ages, with conventional weather information. Carlson (2007) prepared lite imagery in wetlands of Green Bay in Wisconsin, USA. Artigas and an overview of the Triangle Method for estimating surface evapotrans- Yang (2005) used hyperspectral imagery from Airborne Imaging piration and soil moisture from satellite imagery. A review of methods Spectroradiometer for Applications (AISA) in New Jersey, USA to classify using remotely sensed surface temperature data for estimating land sur- marsh condition based on field-captured reflectance spectra of domi- face evaporation is provided by Kalma, McVicar, and McCabe (2008). nant marsh species and seasonal spectra of P. australis. Hestir et al. Continuous reduction of water levels and other man-made and nat- (2008) developed a regional-scale monitoring framework to map wet- ural modifications of wetland hydrology are causing wetland stress and land weeds in the Sacramento–San Joaquin Delta, California USA. They losses in many parts of the world. These hydrologic changes influence focused on the terrestrial riparian weed, perennial pepperweed vegetation species composition, distribution and condition. In worst (Lepidium latifolium) using an airborne hyperspectral imager (HyMAP) cases, this can lead to a drying out of the whole wetland (Chopra, that collects 128 bands in the visible and near-infrared (VNIR; Verma, & Sharma, 2001; Xin, 2004). The hydrological conditions of 0.45–1.50 μm) through the shortwave infrared (SWIR; 1.50–2.5 μm), emergent wetland vegetation have been explored with the help of at bandwidths from 10 nm in the VNIR to 15–20 nm in the SWIR. The remotely sensed biophysical data, such as surface temperature and veg- spatial resolution of the data was 3 m, with a swath width of 1.5 km. etation indices (NDVI). Using a digital elevation model and regression At a field scale Sonnentag et al. (2011) used multispectral webcam models, a relation between surface temperature and water stress has imagery to track perennial pepperweed in the Sacramento Delta in been established (Banks, Paylor, & Hughes, 1996). Results show that California. The success of these studies indicates that HyspIRI has the surface temperature and NDVI can be used for a better understanding potential to be useful at mapping some invasive species. HyspIRI's of hydrological conditions of wetlands (Banks et al., 1996; Bendjoudi, repeat visits should afford the opportunity to monitor spread of those Weng, Guérin, & Pastre, 2002; Petropoulos et al., 2009). MODIS daily species over time. and 8-day composite products have been used effectively to map floodplain and wetland inundation extent corresponding to peak flows of sig- 4. HyspIRI and wetland hydrology and hydro-meteorology nificant flood events (Chen, Huang, Ticehurst, Merrin, & Thew, 2013). In comparison, HyspIRI's higher spatial resolution should provide even Evapotranspiration is an important variable in water and energy more accurate information. balances of the earth's surface. Understanding the distribution of evapo- Satellite data have been used to improve rainfall mapping and to transpiration is a key factor in hydrology, climatology, agronomy, monitor relationships between rainfall and vegetation responses. and ecological studies. On a global scale, about 64% of precipitation on Vegetation and inundation dynamics and aspects of water quality of wet- the continents is evapotranspired. Of this amount, about 97% is lands have been monitored in many parts of the tropics and elsewhere K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217 213 by using multi-spectral, thermal and radar imagery (Meijerink, 2002). To water targets. However, these environments are still subject to sun study the dynamics of regional vegetation responses in southern Africa glint (Hochberg et al., 2011) when observed, which presents a chal- to the dry El Nino years and intervening wetter years, Kogan (1998) lenge for most aquatic remote sensing capabilities along coasts. For used a Vegetation and Temperature condition index (VT) based on coastal emergent vegetation, the issue of glint becomes much more NDVI values after radiometric corrections and the brightness tempera- complex. For example, in salt marshes, tidal emergent vegetation is typ- ture of the surface based on the thermal channel of the Advanced Very ically erectophile, with small ponds and channels interspersed. These High Resolution Radiometer (AVHRR). Effects of transpiration of healthy features are typically ranged in spatial scale from fractions of a meter vegetation are included by the temperature component of the index. to tens of meters. In these systems, glint would undoubtedly contribute Kogan (1998) found that it took 5 to 6 weeks for the VT index to decrease to the remotely sensed signal of the HyspIRI instrument with its 30 m to a level that indicates severe stress (VT = 10–20), whereas with ade- nadir pixel size. Naturally, the effect would vary with sun and view an- quate moisture, the index is around 60. The numerical value of the VT gles, being more intense at low latitudes and growing worse with close- can be used for multi-temporal studies and for comparing values via ness to the summer solstice. Measurement and modeling capabilities the HyspIRI sensors in different parts of the world. for glint in wetlands lag those for shallow and deep oceans. Suitable Natural vegetation patterns, as observed on images, can also be re- models or measurement techniques have yet to be developed to quan- lated to groundwater occurrences and groundwater flow systems tify this effect because the relationship between surface roughness and (Meijerink, 2002). Locating ground-water discharge zones in bodies of various environmental factors are unknown (Turpie, 2012). surface water can provide information about the ground-water flow It is possible to demonstrate glint effects from multi-angle satellite system in wetlands and about the potential transport of contaminants. images acquired by CHRIS/Proba. Fig. 5 illustrates an example of glint This information can aid in the design and emplacement of ground- in the wetlands on remote sensing reflectance observations. With a water monitoring networks and determining remediation techniques. nominal view zenith angle of 0°, glint is visually apparent in water bod- Knowing the areal extent of ground-water discharge to surface water ies among the vegetated areas (Fig. 5a). This glint is likely caused by can also be useful for selecting sampling sites and for estimating the en- capillary waves patterned by the local wind field. With a nominal vironmental effects of contaminant migration (Klemas, 2011, 2013b; view zenith angle of 55°, glint is much less visually apparent (Fig. 5b). Thomson & Nielsen, 1980). Thermal infrared imaging is an effective These visual glint patterns are supported by sample spectra. For the method for assessing large areas and getting information about specific same region of wetland, the 0° nominal view angle spectra (Fig. 6a) locations of ground-water discharge since the ground-water usually has have higher reflectance values and are more variable than the 55° a different temperature from the background water (Byers & Chmura, view angle spectra (Fig. 6b). The bulk of the vegetation signal, seen be- 2014; Thomson & Nielsen, 1980; Xin, 2004). Simulated flooding pat- tween the 1st and 3rd quartiles, shows an increase in visible reflectance terns obtained from coupled surface water-groundwater models have with increased glint and a slight decrease in the NIR, which is possibly been compared to patterns derived from satellite multispectral, thermal an effect of the canopy bidirectional distribution function of a largely infrared and radar data that provided such model inputs as topography, erectophile canopy. aquifer thickness, channel positions, evapotranspiration and precipita- Although using conventional techniques to separate glint from the tion (Milzow, Kgotlhang, Kinzelbach, Meier, & Bauer-Gottwein, 2009). vegetation spectral signal may appear to be challenging, the effect of glint on the relatively stronger signal from subaerial vegetation is less 5. Challenges than the effect on remote sensing retrievals of water column and benthic communities. In addition, wetland vegetation cover tends to reduce sur- Although darker than most terrestrial vegetation remote sensing face roughness from wind, thus the range of angles that are influenced by targets, wetlands are more reflective than most non-turbid, open sun glint are reduced in comparison to ocean water (Vanderbilt et al.,

Fig. 5. Multi-angle CHRIS/Proba images over Maryland marsh systems. Images are cropped and co-registered and pertain to the single-line box in Fig. 2. The 872.1, 551.3, and 490.2 nm bands are used for the red, green, and blue channels, respectively. (a) At 0° nominal view zenith angle glint is visually apparent on water bodies interspersed among subaerial vegetation. (b) At 55° nominal view zenith angle glint is much less apparent. Boxes cover same ground area in (a) and (b). This region is extracted for statistics shown in Fig. 6. 214 K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217

For instance, some areas covered by coastal wetlands may be too small and patchy to be resolved by HyspIRI or any other medium resolution satellite and may require ancillary data. Furthermore, the floristic diversity of some higher-elevation wetlands can have a more mixed vegetation cover, producing a more complex, composite spectral signature than most saltwater wetlands. Mapping of wetland vegetation with high spectral heterogeneity among plant communities caused by variation in background reflectance from water, submerged aquatic vegetation, mud substrate, and dead vegetation could benefit from the use of linear spectral unmixing techniques, such as Multiple Endmember Spectral Mixture Analysis (MESMA) (Roberts et al., 1998). MESMA models measure spectra as linear combination of pure spectra, called endmembers, and allows the types and number of endmembers to vary on a per pixel basis. This method enables vegetation to be characterized by a unique set of endmembers and their fractions. MESMA can improve vegetation mapping even with high-resolution imagery. In a recent application of MESMA in the Sacramento–San Joaquin River Delta, a freshwater marsh, with a higher degree of floristic diversity than typically observed under high saline conditions, was mapped with 2-meter resolution World View-2 imagery. This enabled the accurate classification of green emergent vegetation (Schoenoplectus acutus and Typha spp.), dead standing vegetation, Salix sp., floating aquatic vegetation, and non-vegetated areas (e.g., open water or paved areas) (Byrd et al., 2014), all of which were varying at much finer scales than the sensor's pixel size. The performance of linear unmixing techniques may be limited, in part, by any nonlinear mixing of spectral signatures along the optical path of the sensor. Unless the canopy is very dense, there will be small gaps exposing moist soil or water, which could “contaminate” the pure spectral reflectance signatures of the vegetation. However, light reflected from the canopy substrate and understory must propagate through the canopy above and are subject absorption and multi- scatter processes. This may necessitate the application of canopy reflec- Fig. 6. Spectra extracted from regions highlighted by boxes in Fig. 5. (a) At 0° nominal view tance models or empirical models of vegetation communities, not just zenith angle, glint produces very high values across the spectrum, evidenced by the plant species endmembers for individual plants or leaves. In addition, maximum spectral curve. (b) At 55° nominal view zenith angle, the glint effect is greatly the influence of a varying submerged substrate (e.g., water depth reduced. changing with tides) to the optical properties of any inundated canopies will present additional challenges for some of these techniques. In such cases, a priori information about canopy and substrate conditions the 2002). The 11 AM equatorial crossing of the spacecraft and 4° tilt of the maybe necessary in spectral signature analysis, thus increasing HyspIRI instrument away from the subsolar point should greatly reduce uncertainty. Further sensitivity analysis is recommended to determine the effect of sun glint in these coastal and inland wetlands. the degree of influence variation in substrate water depth variation On the other hand, the consistent satellite acquisition time will not has on canopy spectral signatures (Turpie, 2012, in press). allow for control of tide stage, and differences in tidal inundation Finally, the hydro-meteorological data for many of the world's fresh- when the imagery is captured can influence efforts to map wetland ex- water marshes is still quite inadequate (Jackson, 1985; Mohamed et al., tent and land change (Allen, Couvillion, & Barras, 2012). In addition, the 2004). The areal size of these marshes, the evaporation rates and their optical properties of an inundated canopy can present new challenges influence on the micro and meso climate are still unresolved questions for some of the traditional dry-land remote sensing techniques, such of their hydrology (Carlson, 2007; Rivas & Caselles, 2004) The combina- as red edge position (REP) based methods and other such spectral anal- tion of HyspIRI's thermal infrared radiometer and imaging spectrometer yses (Turpie, 2012, in press). This is driven by the spectral absorption will be effective for observing these hydrologic conditions of wetlands, properties of the aquatic substrate and the fact that the inundated back- including water levels and other hydro-meteorological properties. ground is much less isotropic than dry soil. To date, there have been no true hyperspectral instruments with 6. Conclusion global coverage of coastal regions at 30 m resolution or better. This problem may be remedied as HyspIRI and other advanced satellites According to current plans, HyspIRI will be the only hyperspectral are launched. However, wetlands can exhibit considerable temporal imager supported by the USA with 30 m spatial resolution, which is nec- variability and spatial complexity, as demonstrated in Figs. 1 and 2. essary to observe fine-scale structure in coastal wetlands. It will also be Some key features, such as peak aboveground biomass, are often the only mission in world that would regularly map the wetlands glob- found along channel edges (Schile, Callaway, Parker, & Vasey, 2011). ally. HyspIRI will also uniquely provide contemporaneous observation Thus some finer scale wetland observations are more commensurate in the thermal infrared. Although orbiting, multi-spectral imagers with with higher spatial and temporal resolution imagery, such as those moderate and high spatial resolution have been useful in observing provided by airborne and high spatial resolution satellite sensors. How- wetlands, HyspIRI's combination of hyperspectral imagery with con- ever, at 30 m resolution, spectral mixture analysis could still be used to temporaneous thermal bands would provide greatly enhanced informa- accurately map wetland extent and features masked by mixed pixels at tion. Hyperspectral imagery could offer improved classification of the land/water edge (Hestir et al., 2008; Michishita, Jiang, Gong, & Xu, wetland vegetation functional types and perhaps discriminate species 2012; Rosso et al., 2005). or produce maps of zonation in salt and brackish marshes or dense K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217 215 mangroves. Hyperspectral imagery could also offer the estimation of Barbier, E.B., Hacker, S.D., Kennedy, C., Koch, E.W., Stier, A.C., & Silliman, B.R. (2011). The value of estuarine and coastal ecosystem services. Ecological Monographs, 81(2), biochemical and biophysical parameters of wetland vegetation. 169–193. http://dx.doi.org/10.1890/10-1510.1. HyspIRI's radiometric performance will also provide opportunities to Barbier, E.B., Koch, E.W., Silliman, B.R., Hacker, S.D., Wolanski, E., Primavera, J., et al. observe the optical properties of adjacent water bodies, which could (2008). Coastal ecosystem-based management with nonlinear ecological functions and values. Science, 319,321–323. help support studies of how wetlands tie to marine ecosystems, biogeo- Barducci, A., Guzzi, D., Marcoionni, P., & Pippi, I. (2009). Aerospace wetland monitoring by chemical cycling, and water quality. HyspIRI's thermal infrared and hyperspectral imaging sensors: A case study in the coastal zone of San Rossore natu- hyperspectral sensors will be able to contemporaneously observe hy- ral park. Journal of Environmental Management, 90(7), 2278–2286. drologic and hydro-meteorological processes, including ground water Barnsley, M.J., Settle, J.J., Cutter, M.A., Lobb, D.R., & Teston, F. (2004). The Proba/CHRIS Mission: A low-cost smallsat for hyperspectral multiangle observations of the earth surface and seepage, nitrogen load, water quality, and evapotranspiration (Banks atmosphere. IEEE Transactions on Geoscience and Remote Sensing, 42(7), 1512–1520. et al., 1996; Bendjoudi et al., 2002; Byers & Chmura, 2014; Chen et al., Batzer, D.P., & Baldwin, A.H. (Eds.). (2012). Wetland habitats of North America: Ecology and 2002; Meijerink, 2002; Moffett, 2010; Mohamed et al., 2004; Xin, conservation concerns. Berkeley and Los Angeles: University of California Press (389 pp.). Bendjoudi, H., Weng, P., Guérin, R., & Pastre, J. (2002). Riparian wetlands of the middle 2004). Along with some canopy structure from hyperspectral data reach of the Seine river (France): Historical development, investigation and present (Fan et al., 2010), a more complete model of wetland energy and hydrologic functioning. A case study. Journal of Hydrology, 263(1–4), 131–155. water exchanges with the atmosphere may be obtained. Contempora- http://dx.doi.org/10.1016/S0022-1694(02)00056-2. Blasco, F., Aizpuru, M., & Din Ndongo, D. (2005). Mangroves, remote sensing. In M.L. neous observations of water surface temperature could provide some Schwartz (Ed.), Encyclopedia of coastal science (pp. 614–617). Dordrecht, The insight into coastal freshwater discharges and hence yield further infor- Netherlands: Springer. mation tying wetland processes to open water systems. Blasco, F., Aizpuru, M., & Gers, C. (2001). Depletion of the mangroves of Continental Asia. Wetlands Ecology and Management, 9,245–256. The HyspIRI mission will operationally generate publicly available Brando, V.E., & Dekker, A.G. (2003). Satellite hyperspectral remote sensing for estimating data products that cover the entire globe. Other higher level, derived at- estuarine and coastal water quality. IEEE Transactions on Geoscience and Remote tributes and maps used in wetland research and applications are cur- Sensing, 41(6), 1378–1387. http://dx.doi.org/10.1109/TGRS.2003.812907. Bridgham, S.D., Megonigal, J.P., Keller, J.K., Bliss, N.B., & Trettin, C. (2006). The carbon bal- rently expected to be produced by the research community. Expressed ance of North American wetlands. Wetlands, 26,889–916. community awareness and subsequent interest in using and developing Bromberg-Gedan, K., Silliman, B.R., & Bertness, M.D. (2009). Centuries of human-driven these specialized observations should guide NASA in allocating funding change in salt marsh ecosystems. Annual Review of Marine Science, 1(1), 117–141. and resources to higher level data product generation and distribution, Byers, S.E., & Chmura, G.L. (2014). Observations on shallow subsurface hydrology at Bay of Fundy macrotidal salt marshes. Journal of Coastal Research, 30(5), 1006–1016. http:// as appropriate. To encourage that appropriate support is provided, it is dx.doi.org/10.2112/JCOASTRES-D-12-00167.1. imperative that wetland scientists voice their needs early to the agency Byrd, K.B., O'Connell, J.L., Di Tommaso, S., & Kelly, M. (2014). Evaluation of sensor types as the mission development proceeds. This is likewise important in and environmental controls on mapping biomass of coastal marsh emergent vegetation. Remote Sensing of Environment, 149,166–180. guiding the NASA and other funding agencies in identifying open ques- Carlson, T. (2007). An overview of the “triangle method” for estimating surface evapo- tions and understanding the challenges, such as those described in this transpiration and soil moisture from satellite imagery. Sensors, 7(8), 1612–1629. paper, when allocating research resources and developing funding http://dx.doi.org/10.3390/s7081612. Cebrian, J. (2002). Variability and control of carbon consumption, export, and accumula- opportunities. tion in marine communities. Limnology and Oceanography, 47(1), 11–22. Chapman, V.J. (1960). Salt marshes and salt deserts of the world. London: Leon Hill (392 pp.). Acknowledgments Chen, Y., Huang, C., Ticehurst, C., Merrin, L., & Thew, P. (2013). An evaluation of MODIS daily and 8-day composite products for floodplain and wetland inundation mapping. Wetlands, 33(5), 823–835. http://dx.doi.org/10.1007/s13157-013-0439-4. We appreciate useful input from the HyspIRI Aquatic Studies Group Chen, J. -H., Kan, C. -E., Tan, C. -H., & Shih, S. -F. (2002). Use of spectral information for (HASG), which is sponsored by NASA's Goddard Space Flight Center, wetland evapotranspiration assessment. Agricultural Water Management, 55(3), 239–248. http://dx.doi.org/10.1016/S0378-3774(01)00143-3. and now part of AquaRS, a new organization facilitating a community Chopra, R., Verma, V.K., & Sharma, P.K. (2001). Mapping, monitoring and conservation of of practice for aquatic remote sensing in coastal and inland waters. Harike wetland ecosystem, Punjab, India, through remote sensing. International Journal Data from the Surrey Satellite Technology Ltd CHRIS instrument aboard of Remote Sensing, 22(1), 89–98. http://dx.doi.org/10.1080/014311601750038866. Christian, B., & Krishnayya, N. (2009). Classification of tropical trees growing in a sanctu- the European Space Agency (ESA) PROBA platform were provided by ary using Hyperion (EO-1) and SAM algorithm. Current Science, 96(12), 1601–1607. the ESA in the framework of the Cat-1 project “Validation of a marsh Church, J.A., & White, N.J. (2006). A 20th century acceleration in global sea-level rise. canopy radiative transfer model” (ID 4258). Geophysical Research Letters, 33(1), L01602. http://dx.doi.org/10.1029/2005GL024826. Dahl, T.E., & Stedman, S.M. (2013). Status and trends of wetlands in the coastal watersheds of the Conterminous United States 2004 to 2009. U.S. Department of the Interior, Fish References and Wildlife Service and National Oceanic and Atmospheric. Administration, National Marine Fisheries Service (46 pp. http://www.fws.gov/wetlands/Documents/Status- Adam, P. (1990). Saltmarsh ecology. Cambridge, England: Cambridge University Press. and-Trends-of-Wetlands-In-the-Coastal-Watersheds-of-the-Conterminous-US- Adam, E., & Mutanga, O. (2009). Spectral discrimination of papyrus vegetation (Cyperus 2004-to-2009.pdf. Date accessed: January 23, 2014). papyrus L.) in swamp wetlands using field spectrometry. ISPRS Journal of Day, J. W., Jr., Kemp, W. M., Yañez-Arancibia, A., & Crump, B. C. (2012). Estuarine ecology. Photogrammetry and Remote Sensing, 64(6), 612–620. John Wiley & Sons; New York ISBN-13: 978-0471755678. Adam, E., Mutanga, O., & Rugege, D. (2010). Multispectral and hyperspectral remote sens- Devred, E., Turpie, K.R., Moses, W., Klemas, V.V., Moisan, T., Babin, M., et al. (2013). Future ing for identification and mapping of wetland vegetation: A review. Wetlands Ecology retrievals of water column bio-optical properties using the Hyperspectral Infrared and Management, 18(3), 281–296. http://dx.doi.org/10.1007/s11273-009-9169-z. Imager (HyspIRI). Remote Sensing, 5(12), 6812–6837. Allen, Y.C., Couvillion, B.R., & Barras, J.A. (2012). Using multitemporal remote sensing im- Erickson, J.E., Megonigal, J.P., Peresta, G., & Drake, B.G. (2007). Salinity and sea level agery and inundation measures to improve land change estimates in coastal mediate elevated CO2 effects on C3/C4 plant interactions and tissue nitrogen in a wetlands. Estuaries and Coasts, 35,190–200. Chesapeake Bay tidal wetland. Global Change Biology, 13(1), 202–215. Alongi,D.M.(2002).Presentstateandfuture of the world's mangrove forests. Environmental Ewel, K., Twilley, R., & Ong, J.I.N. (1998). Different kinds of mangrove forests provide dif- Conservation, 29(03), 331–349. http://dx.doi.org/10.1017/S0376892902000231. ferent goods and services. Global Ecology and Biogeography Letters, 7,83–94. Andrew, M., & Ustin, S. (2008). The role of environmental context in mapping invasive Fan, W.J., Xu, X.R., Liu, X.C., & Cui, Y.K. (2010). Accurate LAI retrieval method based on plants with hyperspectral image data. Remote Sensing of Environment, 112(12), PROBA/CHRIS data. Hydrology and Earth System Sciences, 14, 1499–1507. http://dx. 4301–4317. doi.org/10.5194/hess-14-1499-2010. Artigas, F.J., & Yang, J.S. (2005). Hyperspectral remote sensing of marsh species and plant Fearns, P.R.C., Klonowski, W., Babcock, R.C., England, P., & Phillips, J. (2011). Shallow water vigour gradient in the New Jersey Meadowlands. International Journal of Remote substrate mapping using hyperspectral remote sensing. Continental Shelf Research, Sensing, 26(23), 5209–5220. http://dx.doi.org/10.1080/01431160500218952. 31(12), 1249–1259. http://dx.doi.org/10.1016/j.csr.2011.04.005. Artigas, Francisco J., & Yang, J. (2006). Spectral discrimination of marsh vegetation types Filippi, A.M., & Jensen, J.R. (2006). Fuzzy learning vector quantization for hyperspectral in the New Jersey Meadowlands, USA. Wetlands, 26(1), 271–277. http://dx.doi.org/ coastal vegetation classification. Remote Sensing of Environment, 100(4), 512–530. 10.1672/0277-5212(2006)26[271:SDOMVT]2.0.CO;2. http://dx.doi.org/10.1016/j.rse.2005.11.007. Banks, W.S.L., Paylor, R.L., & Hughes, W.B. (1996). Using thermal-infrared imagery to de- Friess, D.A., & Webb, E.L. (2014). Variability in mangrove change estimates and implica- lineate ground-water discharged. Ground Water, 34(3), 434–443. http://dx.doi.org/ tions for the assessment of ecosystem service provision. Global Ecology and 10.1111/j.1745-6584.1996.tb02024.x. Biogeography, 23(7), 715–725. http://dx.doi.org/10.1111/geb.12140. Barbier, E. M. (2011). Wetlands as natural assets. Hydrological Sciences Journal - Journal Gao, J. (1998). A hybrid method toward accurate mapping of mangroves in a marginal des Sciences Hydrologiques, 56(8), 1360–1373. http://dx.doi.org/10.1080/02626667. habitat from SPOT multispectral data. International Journal of Remote Sensing, 2011.629787. 19(10), 1887–1899. http://dx.doi.org/10.1080/014311698215045. 216 K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217

Ghilain, N., Arboleda, A., Sepulcre-Canto, G., Batelaan, O., Ardo, J., & Gellens- Lunetta, R.S., & Balogh, M.E. (1999). Application of multi-temporal Landsat 5 TM imagery Meulenberghs, F. (2011). Improving evapotranspiration in land surface models for wetland identification. Photogrammetric Engineering and Remote Sensing, 65, using biophysical parameters derived from MSG/SEVIRI satellite. Hydrology and 1303–1310. Earth System Sciences, 16(8), 2567–2583. Lyon, J.G., & McCarthy, J. (1995). Wetland and environmental applications of GIS. New York, Gilmore, M.S., Civco, D.L., Wilson, E.H., Barrett, N., Prisloe, S., Hurd, J.D., et al. (2010). Re- U.S.A.: Lewis Publishers (400 pp.). mote sensing and in situ measurements for delineation and assessment of coastal Marshall, M., Tu, K., Funk, C., Michaelsen, J., Williams, P., Williams, C., et al. (2013). Im- marshes and their constituent species. In J. Wang (Ed.), Remote sensing of coastal proving operational land surface model canopy evapotranspiration in Africa using a environment. Boca Raton, Florida: CRC Press. direct remote sensing approach. Hydrology and Earth System Sciences, 17(3), Gilmore, M.S., Wilson, E.H., Barrett, N., Civco, D.L., Prisloe, S., Hurd, J.D., et al. (2008). Inte- 1079–1091. http://dx.doi.org/10.5194/hess-17-1079-2013. grating multi-temporal spectral and structural information to map wetland vegeta- McInnes, K., Walsh, K., Hubbert, G., & Beer, T. (2003). Impact of sea-level rise and storm tion in a lower Connecticut River tidal marsh. Remote Sensing of Environment, surges on a coastal community. Natural Hazards, 30, 187–207. http://dx.doi.org/10. 112(11), 4048–4060. http://dx.doi.org/10.1016/j.rse.2008.05.020. 1023/A1026118417752. Green, R. (2015). Personal communication via e-mail from Dr. Robert Green, JPL — February Meijerink, A. (2002). Satellite eco-hydrology: A review. Tropical Ecology, 43(1), 91–106. 26, 2015. Mendelssohn, I., McKee, K., & Kong, T. (2001). A comparison of physiological indicators of Gunawardena, M., & Rowan, J.S. (2005). Economic valuation of a mangrove ecosystem sublethal cadmium stress in wetland plants. Environmental and Experimental Botany, threatened by shrimp aquaculture in Sri Lanka. Environmental Management, 36(4), 46(3), 263–275. http://dx.doi.org/10.1016/S0098-8472(01)00106-X. 535–550. http://dx.doi.org/10.1007/s00267-003-0286-9. Michishita, R., Jiang, Z., Gong, P., & Xu, B. (2012). Bi-scale analysis of multitemporal land Hestir, E.L., Khanna, S., Andrew, M.E., Santos, M.J., Viers, J.H., Greenberg, J.A., et al. (2008). cover fractions for wetland vegetation mapping. ISPRS Journal of Photogrammetry Identification of invasive vegetation using hyperspectral remote sensing in the Cali- and Remote Sensing, 72,1–15. fornia Delta ecosystem. Remote Sensing of Environment, 112(11), 4034–4047. http:// Miller, R.L., & Fujii, R. (2010). Plant community, primary productivity, and environmental dx.doi.org/10.1016/j.rse.2008.01.022. conditions following wetland re-establishment in the Sacramento–San Joaquin Delta, Heumann, B.W. (2011). Satellite remote sensing of mangrove forests: Recent advances California. Wetlands Ecology and Management, 18(1), 1–16. http://dx.doi.org/10.1007/ and future opportunities. Progress in Physical Geography, 35(1), 87–108. http://dx. s11273-009-9143-9. doi.org/10.1177/0309133310385371. Milzow, C., Kgotlhang, L., Kinzelbach, W., Meier, P., & Bauer-Gottwein, P. (2009). The role Hirano, A., Madden, M., & Welch, R. (2003). Hyperspectral image data for mapping of remote sensing in hydrological modelling of the Okavango Delta, Botswana. Journal wetland vegetation. Wetlands, 23(2), 436–448. http://dx.doi.org/10.1672/18- of Environmental Management, 90(7), 2252–2260. http://dx.doi.org/10.1016/j. 20. jenvman.2007.06.032. Hochberg, E.J., Bruce, C.F., Green, R.O., Oaida, B.V., Muller-Karger, F.E., Mobley, C.D., et al. Mishra, D.R., Cho, H.J., Ghosh, S., Fox, A., Downs, C., Merani, P.B.T., et al. (2012). Post-spill (2011). HyspIRI Sun Glint Report. JPL Publication 11-4. Pasadena, CA: NASA JPL state of the marsh: Remote estimation of the ecological impact of the Gulf of Mexico (67 pp.). oil spill on Louisiana Salt Marshes. Remote Sensing of Environment, 118,176–185. Jackson, R. (1985). Evaluating evapotranspiration at local and regional scales. Proceedings Moffett, K.B. (2010). Groundwater–vegetation–atmosphere interactions in an intertidal salt of the IEEE, 73(6), 1086–1096. http://dx.doi.org/10.1109/PROC.1985.13239. marsh. (Ph.D. dissertation) Stanford, CA: Department of Environmental Earth System Jensen, J.R., Coombs, C., Porter, D., Jones, B., Schill, S., & White, D. (1998). Extraction of Science, Stanford University. smooth cordgrass (Spartina alterniflora) biomass and leaf area index parameters Mohamed, Y.A., Bastiaanssen, W.G.M., & Savenije, H.H.G. (2004). Spatial variability of from high resolution imagery. Geocarto International, 13(4), 25–34. http://dx.doi. evaporation and moisture storage in the swamps of the upper Nile studied by remote org/10.1080/10106049809354661. sensing techniques. Journal of Hydrology, 289(1–4), 145–164. http://dx.doi.org/10. Jensen, J.R., Hodgson, M.E., Christensen, E., Mackey, H.E., Tinney, L.R., & Sharitz, R. (1986). 1016/j.jhydrol.2003.11.038. Remote-sensing inland wetlands — a multispectral approach. Photogrammetric Mutanga, O., Adam, E., & Cho, M.A. (2012). High density biomass estimation for wetland Engineering and Remote Sensing, 52(1), 87–100. vegetation using WorldView-2 imagery and random forest regression algorithm. Jensen, R., Mausel, P., Dias, N., Gonser, R., Yang, C., Everitt, J., et al. (2007). Spectral analysis International Journal of Applied Earth Observation and Geoinformation, 18,399–406. of coastal vegetation and land cover using AISA+ hyperspectral data. Geocarto Mutanga, O., & Skidmore, A.K. (2004). Narrow band vegetation indices overcome the sat- International, 22(1), 17–28. http://dx.doi.org/10.1080/10106040701204354. uration problem in biomass estimation. International Journal of Remote Sensing, 25, Jollineau, M.Y., & Howarth, P.J. (2008). Mapping an inland wetland complex using 3999–4014. hyperspectral imagery. International Journal of Remote Sensing, 29(12), 3609–3631. NOAA Office for Coastal Management (2015). NOAA Coastal Change Analysis Program (C- http://dx.doi.org/10.1080/01431160701469099. CAP) Regional Land Cover Database. Data collected 1995–present. Data accessed at Judd, C., Steinberg, S., Shaughnessy, F., & Crawford, G. (2007). Mapping salt marsh vegetation www.coast.noaa.gov/digitalcoast/data/ccapregional (Charleston, SC) using aerial hyperspectral imagery and linear unmixing in Humboldt Bay, California. Nungesser, M.K. (2011). Reading the landscape: Temporal and spatial changes in a pat- Wetlands, 27(4), 1144–1152. http://dx.doi.org/10.1672/0277-5212(2007)27[1144: terned peatland. Wetlands Ecology and Management, 19,475–493. MSMVUA]2.0.CO;2. O'Connell, J.L., Byrd, K.B., & Kelly, M. (2014). Remotely-sensed indicators of N-related bio- Kalma, J.D., McVicar, T.R., & McCabe, M.F. (2008). Estimating land surface evaporation: A mass allocation in Schoenoplectus acutus. PLoS ONE, 9, e90870. review of methods using remotely sensed surface temperature data. Surveys in Odum, E.P. (1993). Ecology and our endangered life-support systems (2nd edition ). Geophysics, 29,421–469. Sunderland, Massachusetts, U.S.A.: Sinauer Associates, Inc (320 pp.). Kelly, M., & Tuxen, K. (2009). Remote sensing support for tidal wetland vegetation Re- Ozesmi, S.L., & Bauer, M.E. (2002). Satellite remote sensing of wetlands. Wetlands Ecology search and management. In X. Yang (Ed.), Remote sensing and geospatial technologies and Management, 10,381–402. http://dx.doi.org/10.1023/A:1020908432489. for coastal ecosystem assessment and management (pp. 341–364). Berlin, Germany: Papes, M., Tupayachi, R., Martínez, P., Peterson, A.T., & Powell, G.V.N. (2010). Using Springer-Verlag. hyperspectral satellite imagery for regional inventories: A test with tropical emergent Kelly, M., Tuxen, K.A., & Stralberg, D. (2011). Mapping changes to vegetation pattern in a trees in the Amazon Basin. Journal of Vegetation Science, 21(2), 342–354. http://dx.doi. restoring wetland: Finding pattern metrics that are consistent across spatial scale and org/10.1111/j.1654-1103.2009.01147.x. time. Ecological Indicators, 11(2), 263–273. http://dx.doi.org/10.1016/j.ecolind.2010. Pendleton, L., Donato, D.C., Murray, B.C., Crooks, S., Jenkins, W.A., Sifleet, S., et al. (2012). 05.003. Estimating global “Blue Carbon” emissions from conversion and degradation of veg- Klemas, V. (2011). Remote sensing of wetlands: Case studies comparing practical tech- etated coastal ecosystems. PLoS ONE, 7, e43542. niques. Journal of Coastal Research, 27,418–427. http://dx.doi.org/10.2112/ Pengra, B.W., Johnston, C.A., & Loveland, T.R. (2007). Mapping an invasive plant, JCOASTRES-D-10-00174.1. Phragmites australis, in coastal wetlands using the EO-1 Hyperion hyperspectral sen- Klemas, V. (2013a). Remote sensing of coastal wetland biomass: An overview. Journal of sor. Remote Sensing of Environment, 108(1), 74–81. http://dx.doi.org/10.1016/j.rse. Coastal Research, 290, 1016–1028. http://dx.doi.org/10.2112/JCOASTRES-D-12- 2006.11.002. 00237.1. Petropoulos, G., Carlson, T.N., Wooster, M.J., & Islam, S. (2009). A review of Ts/VI remote Klemas, V. (2013b). Remote sensing of emergent and submerged wetlands: An overview. sensing based methods for the retrieval of land surface energy fluxes and soil surface International Journal of Remote Sensing, 34(18), 6286–6320. http://dx.doi.org/10. moisture. Progress in Physical Geography, 33(2), 224–250. http://dx.doi.org/10.1177/ 1080/01431161.2013.800656. 0309133309338997. Koedsin, W., & Vaiphasa, C. (2013). Discrimination of tropical mangroves at the species Pinet, P.R. (2009). Invitation to oceanography (5th edition ). Sudbury, Massachussetts, level with EO-1 Hyperion data. Remote Sensing, 5, 3562–3582. http://dx.doi.org/10. U.S.A.: Jones and Bartlett (620 pp.). 3390/rs5073562. Ramsey, E., & Rangoonwala, A. (2005). Leaf optical property changes associated with the Kogan, F.N. (1998). A typical pattern of vegetation conditions in southern Africa during El occurrence of Spartina alterniflora dieback in coastal Louisiana related to remote sens- Nino years detected from AVHRR data using three-channel numerical index. ing mapping. Photogrammetric Engineering and Remote Sensing, 71(3), 299–311. International Journal of Remote Sensing, 19(18), 3688–3694. http://dx.doi.org/10. Ramsey, E., & Rangoonwala, A. (2006). Canopy reflectance related to marsh dieback onset 1080/014311698213902. and progression in coastal Louisiana. Photogrammetric Engineering and Remote Kovacs, J.M., Wang, J., & Blanco-Correa, M. (2001). Mapping disturbances in a mangrove Sensing, 72(6), 641–652. forest using multi-date Landsat TM imagery. Environmental Management, 27(5), Rivas, P., & Caselles, V. (2004). A simplified equation to estimate spatial reference evapo- 763–776. http://dx.doi.org/10.1007/s002670010186. ration from remote sensing-based surface temperature and local meteorological data. Lesser, M.P., & Mobley, C.D. (2007). Bathymetry, water optical properties, and benthic Remote Sensing of Environment, 93,68–76. classification of coral reefs using hyperspectral remote sensing imagery. Coral Reefs, Roberts, D.A., Gardner, M., Church, R., Ustin, S., Scheer, G., & Green, R.O. (1998). Mapping 26(4), 819–829. http://dx.doi.org/10.1007/s00338-007-0271-5. chaparral in the Santa Monica Mountains using multiple endmember spectral mix- Li, L., Ustin, S.L., & Lay, M. (2005). Application of multiple endmember spectral mixture ture models. Remote Sensing of Environment, 65,267–279. analysis (MESMA) to AVIRIS imagery for coastal salt marsh mapping: A case study Rocha, A.V., & Goulden, M.L. (2009). Why is marsh productivity so high? New insights in China Camp, CA, USA. International Journal of Remote Sensing, 26(23), 5193–5207. from eddy covariance and biomass measurements in a Typha marsh. Agricultural http://dx.doi.org/10.1080/01431160500218911. and Forest Meteorology, 149,159–168. K.R. Turpie et al. / Remote Sensing of Environment 167 (2015) 206–217 217

Rosso, P.H., Ustin, S.L., & Hastings, A. (2005). Mapping marshland vegetation of San ent using high spatial resolution imagery. Wetlands Ecology and Management, 19(2), Francisco Bay, California, using hyperspectral data. International Journal of Remote 141–157. http://dx.doi.org/10.1007/s11273-010-9207-x. Sensing, 26(23), 5169–5191. http://dx.doi.org/10.1080/01431160500218770. Underwood, E.C., Mulitsch, M.J., Greenberg, J.A., Whiting, M.L., Ustin, S.L., & Kefauver, Saito, H., Bellan, M., Al-Habshi, A., Aizpuru, M., & Blasco, F. (2003). Mangrove research and S.C. (2006). Mapping invasive aquatic vegetation in the Sacramento–San Joaquin coastal ecosystem studies with SPOT-4 HRVIR and TERRA ASTER in the Arabian Gulf. delta using hyperspectral imagery. Environmental Monitoring and Assessment, International Journal of Remote Sensing, 24(21), 4073–4092. http://dx.doi.org/10. 121(1–3), 47–64. 1080/0143116021000035030. UNEP (2006). Marine and coastal ecosystems and human well-being: A synthesis report Sandholt, I., Rasmussen, K., & Andersen, J. (2002). A simple interpretation of the surface based on findings of the millennium ecosystem assessment. Nairobi, Kenya: UNEP. temperature/vegetation index space for assessment of surface moisture status. USFWS (2014). National Wetlands Inventory website. Washington, D.C.: U.S. Department Remote Sensing of Environment, 79(2–3), 213–224. http://dx.doi.org/10.1016/S0034- of the Interior, Fish and Wildlife Service (Publication Date — 1 Oct 2014 (see http:// 4257(01)00274-7. www.fws.gov/wetlands/)). Schile, L.M., Callaway, J.C., Parker, V.T., & Vasey, M.C. (2011). Salinity and inundation influ- USFWS (2015). National report on the implementation of the Ramsar convention on wet- ence productivity of the halophytic plant Sarcocornia pacifica. Wetlands, 31, lands. (Uruguay). 1165–1174. Ustin, S.L., Roberts, D.A., Gamon, J.A., Asner, G.P., & Green, R.O. (2004). Using imaging Schmidt, K.S., & Skidmore, A.K. (2003). Spectral discrimination of vegetation types in a spectroscopy to study ecosystem processes and properties. BioScience, 54(6), coastal wetland. Remote Sensing of Environment, 85(1), 92–108. http://dx.doi.org/10. 523–534. http://dx.doi.org/10.1641/0006-3568(2004)054[0523:UISTSE]2.0.CO;2. 1016/S0034-4257(02)00196-7. Vaesen, K., Gilliams, S., Nackaerts, K., & Coppin, P. (2001). Ground-measured spectral sig- Schmidt, K.S., Skidmore, A.K., Kloosterman, E., Van Oosten, H., Kumar, L., & Janseen, J. natures as indicators of ground cover and leaf area index: The case of paddy rice. Field (2004). Mapping coastal vegetation using an expert system and hyperspectral imag- Crops Research, 69(1), 13–25. http://dx.doi.org/10.1016/S0378-4290(00)00129-5. ery. Photogrammetric Engineering and Remote Sensing, 70(6), 703–715. Vaiphasa, C., Ongsomwang, S., Vaiphasa, T., & Skidmore, A.K. (2005). Tropical mangrove Seher, S.S., & Tueller, P.T. (1973). Color aerial photos for marshland. Photogrammetric species discrimination using hyperspectral data: A laboratory study. Estuarine, Engineering, 38(6), 489–499. Coastal and Shelf Science, 65(1–2), 371–379. http://dx.doi.org/10.1016/j.ecss.2005. Silvestri, S., Marani, M., & Marani, A. (2003). Hyperspectral remote sensing of salt marsh 06.014. vegetation, morphology and soil topography. Physics and Chemistry of the Earth, Vanderbilt, V.C., Perry, G.L., Livingston, G.P., Ustin, S.L., Barrios, M.C. -D., Beron, F.M., et al. 28(1–3), 15–25. http://dx.doi.org/10.1016/S1474-7065(03)00004-4. (2002). Inundation discriminated using sun glint. IEEE Transactions on Geoscience and Sone, C., Saito, K., & Futakuchi, K. (2009). Comparison of three methods for estimating leaf Remote Sensing, 40(6), 1279–1287. area index of upland rice cultivars. Crop Science, 49(4), 1438–1443. http://dx.doi.org/ Varni, M., Usunoff, E., Weinzettel, P., & Rivas, R. (1999). The groundwater recharge in the 10.2135/cropsci2008.09.0520. Azul aquifer, central Buenos Aires Province, Argentina. Physics and Chemistry of the Sonnentag, O., Detto, M., Vargas, R., Ryu, Y., Runkle, B.R.K., Kelly, M., et al. (2011). Tracking Earth, 24,349–352. the structural and functional development of a perennial pepperweed (Lepidium Wang, Y. (2010). Remote sensing of coastal environments: An overview. In J. Wang (Ed.), latifolium L.) infestation using a multi-year archive of webcam imagery and eddy co- Remote sensing of coastal environments (pp. 1–24). Boca Raton, Florida, U.S.A.: CRC variance measurements. Agricultural and Forest Meteorology, 151(7), 916–926. http:// Press. dx.doi.org/10.1016/j.agrformet.2011.02.011. Wang, F., Huang, J., Tang, Y., & Wang, X. (2007). New vegetation index and its application Terchunian, A., Klemas, V., Segovia, A., Alvarez, A., Vasconez, B., & Guerrero, L. (1986). in estimating leaf area index of rice. Rice Science, 14(3), 195–203. http://dx.doi.org/10. Mangrove mapping in Ecuador: The impact of shrimp pond construction. 1016/S1672-6308(07)60027-4. Environmental Management, 10(3), 345–350. http://dx.doi.org/10.1007/BF01867258. Wang, L., & Sousa, W.P. (2009). Distinguishing mangrove species with laboratory mea- Thomson, K.P.B., & Nielsen, G. (1980). Groundwater discharge detection along the coasts surements of hyperspectral leaf reflectance. International Journal of Remote Sensing, of the Arabian Gulf and the Gulf of Oman using thermal infrared imagery. Proceedings 30(5), 1267–1281. http://dx.doi.org/10.1080/01431160802474014. of the 14th International Symposium on Remote Sensing of Environment, San Jose, Costa Watson, E.B., Oczkowski, A.J., Wigand, C., Hanson, A.R., Davey, E.W., Crosby, S.C., et al. Rica, April 23–30, 1980 (pp. 835–843). (2014). Nutrient enrichment and precipitation changes do not enhance resiliency Tian, Y.C., Yao, X., Yang, J., Cao, W.X., Hannaway, D.B., & Zhu, Y. (2011). Assessing newly of salt marshes to sea level rise in the Northeastern U.S. Climatic Change, 125, developed and published vegetation indices for estimating rice leaf nitrogen concen- 501–509. tration with ground- and space-based hyperspectral reflectance. Field Crops Research, Xavier, A.C., & Vettorazzi, C.A. (2004). Mapping leaf area index through spectral vegeta- 120(2), 299–310. http://dx.doi.org/10.1016/j.fcr.2010.11.002. tion indices in a subtropical watershed. International Journal of Remote Sensing, Tilley, D.R., Ahmed, M., Son, J.H., & Badrinarayanan, H. (2003). Hyperspectral reflectance 25(9), 1661–1672. http://dx.doi.org/10.1080/01431160310001620803. of emergent macrophytes as an indicator of water column ammonia in an Xiao, X., He, L., Salas, W., Li, C., Moore, B., Zhao, R., et al. (2002). Quantitative relationships oligohaline, subtropical marsh. Ecological Engineering, 21(2–3), 153–163. http://dx. between field-measured leaf area index and vegetation index derived from doi.org/10.1016/j.ecoleng.2003.10.004. VEGETATION images for paddy rice fields. International Journal of Remote Sensing, Tomlinson, P.B. (1995). The botany of mangroves. Cambridge tropical biology series. 23(18), 3595–3604. http://dx.doi.org/10.1080/01431160110115799. Cambridge: Cambridge University Press (436 pp. ISBN: 978-0521466752). Xin, S. (2004). Inferring wetland hydrological conditions from remote sensing: A case study of Tornqvist, T.E., Wallace, D.J., Storms, J.E.A., Wallinga, J., van Dam, R.L., Blaauw, M., et al. Lake Baiyang, China. (M.S. thesis) Enschede, The Netherlands: International Institute (2008). Mississippi Delta subsidence primarily caused by compaction of Holocene for Geo-Information Science and Earth Observation. strata. Nature Geoscience, 1,173–176. Yang, C., Everitt, J.H., Fletcher, R.S., Jensen, J.R., & Mausel, P.W. (2009). Evaluating AISA+ Turpie, K.R. (2012). Enhancement of a canopy reflectance model for understanding the spec- hyperspectral imagery for mapping black mangrove along the south Texas gulf coast. ular and spectral effects of an aquatic background in an inundated tidal canopy. (Ph.D. Photogrammetric Engineering and Remote Sensing, 75(4), 425–436. dissertation) College Park, Maryland: Department of Geography, University of Zedler,J.B.,&Kercher,S.(2005).Wetland resources: Status, trends, ecosystem services, and Maryland. restorability. Annual Review of Environment and Resources. Palo Alto: Annual Re- Turpie, K.R. (2013). Explaining the spectral red-edge features of inundated marsh views, 39–74. vegetation. Journal of Coastal Research. http://dx.doi.org/10.2112/JCOASTRES-D-12- Zhao, B., Yan, Y., Guo, H., He, M., Gu, Y., & Li, B. (2009). Monitoring rapid vegetation suc- 00209.1 (in press). cession in estuarine wetland using time series MODIS-based indicators: An applica- Tuxen, K.A., Schile, L.M., Kelly, M., & Siegel, S.W. (2008). Vegetation colonization in a re- tion in the Yangtze River Delta area. Ecological Indicators, 9(2), 346–356. http://dx. storing tidal marsh: A remote sensing approach. Restoration Ecology, 16(2), doi.org/10.1016/j.ecolind.2008.05.009. 313–323. http://dx.doi.org/10.1111/j.1526-100X.2007.00313.x. Zomer, R.J., Trabucco, A., & Ustin, S.L. (2009). Building spectral libraries for wetlands land Tuxen, K., Schile, L., Stralberg, D., Siegel, S., Parker, T., Vasey, M., et al. (2011). Mapping cover classification and hyperspectral remote sensing. Journal of Environmental changes in tidal wetland vegetation composition and pattern across a salinity gradi- Management, 90(7), 2170–2177. http://dx.doi.org/10.1016/j.jenvman.2007.